Abstract
Background: Optimized application of telemonitoring may enhance specialized medical care of patients with heart failure (HF). Our objective was to study methods to minimise the amount of alarms generated by the telemonitoring system, without losing clinically important information.
Methods: We included in a pilot project, 25 clinically stable patients with chronic HF. The data were obtained through the iMediSense™ telemonitoring system (2016, Thales, Hengelo, the Netherlands), clinically supervised by cardiologists and nurse practitioners (NPs). The patients were instructed to conduct measurements at least once daily for 60 days: diastolic blood pressure (dBP), systolic blood pressure (sBP), heart rate (HR) and body weight, also they filled out a HF-symptoms-questionnaire. When measurements exceeded predefined ranges, alarms were generated. NPs were instructed to view the generated alarms and react accordingly. We carried out simulations of different alarm systems: (weight-based) rule of thumb, moving average convergence divergence (MACD) and moving average (MA). The primary outcome of our study was the adjustment of HF medication.
Findings: Based on the symptoms questionnaires, only few adjustments of medication were carried out (n=10). A total of 865 responses were given by NPs to the 1471 generated alarms. The most common response yielded wait-and-see (47%, n=406). The response of the NPs was undocumented by the iMediSense™ system in 45% of the cases (n=391). The interview with NPs revealed that the NP will only respond to when a trend is visible. Both sBP and HR driven alarms were able to detect true positives (TPs). MA simulation driven by both parameters gave the highest amount of TPs, 0.6% (n=9) and lowest amount of FPs, 33.5% (n=293).
Discussion: In stable heart failure patients we found that a moving average algorithm based on systolic blood pressure and heart rate, improved accuracy of alarms generated by the telemonitoring system, in comparison to preset thresholds.
Methods: We included in a pilot project, 25 clinically stable patients with chronic HF. The data were obtained through the iMediSense™ telemonitoring system (2016, Thales, Hengelo, the Netherlands), clinically supervised by cardiologists and nurse practitioners (NPs). The patients were instructed to conduct measurements at least once daily for 60 days: diastolic blood pressure (dBP), systolic blood pressure (sBP), heart rate (HR) and body weight, also they filled out a HF-symptoms-questionnaire. When measurements exceeded predefined ranges, alarms were generated. NPs were instructed to view the generated alarms and react accordingly. We carried out simulations of different alarm systems: (weight-based) rule of thumb, moving average convergence divergence (MACD) and moving average (MA). The primary outcome of our study was the adjustment of HF medication.
Findings: Based on the symptoms questionnaires, only few adjustments of medication were carried out (n=10). A total of 865 responses were given by NPs to the 1471 generated alarms. The most common response yielded wait-and-see (47%, n=406). The response of the NPs was undocumented by the iMediSense™ system in 45% of the cases (n=391). The interview with NPs revealed that the NP will only respond to when a trend is visible. Both sBP and HR driven alarms were able to detect true positives (TPs). MA simulation driven by both parameters gave the highest amount of TPs, 0.6% (n=9) and lowest amount of FPs, 33.5% (n=293).
Discussion: In stable heart failure patients we found that a moving average algorithm based on systolic blood pressure and heart rate, improved accuracy of alarms generated by the telemonitoring system, in comparison to preset thresholds.
| Original language | English |
|---|---|
| Publication status | Published - 1 Jun 2018 |
| Event | 8ste Conference Supporting Health By Technology 2018 - University of Twente, Enschede, Netherlands Duration: 1 Jun 2018 → 1 Jun 2018 Conference number: 8 http://healthbytech.com/ |
Conference
| Conference | 8ste Conference Supporting Health By Technology 2018 |
|---|---|
| Abbreviated title | HealthbyTech 2018 |
| Country | Netherlands |
| City | Enschede |
| Period | 1/06/18 → 1/06/18 |
| Internet address |